FastCount

Motivation: The advent of single cell RNA-seq (scRNA-seq) enables scientists to characterize the transcriptomic response of cells under different conditions and understand expression heterogeneity at single cell level. One of the fundamental steps in scRNA-seq analysis is to summarize raw sequencing reads into a list of gene counts for each individual cell. However, this step remains to be most time-consuming and resource intensive in the analysis workflow due to the large amount of data produced in a scRNA-seq experiment. It is further complicated by the special handling of cell barcodes and unique molecular identifiers (UMIs) information in the read sequences. For example, the gene count summarization of 10X Chromium sequencing by standard Cell Ranger count often takes many hours to finish when running on a computing cluster. Although several alignment-free algorithms have been developed to improve efficiency, their derived gene count suffer from poor concordance with Cell Ranger count and algorithm-specific bias[1].